Fig. 2

Change in probabilities of dropping out of high school and receiving income assistance reported by women with no teen pregnancy compared with those experiencing teen pregnancy.

* Notes: Figure shows marginal effects (ME) and 95% confidence intervals (CI) calculated based on four GEE models with binary distribution, binomial link, and exchangeable correlation. Model 1 (M1) regressed the teen pregnancy indicators (none, live birth, stillbirth/miscarriage, abortion) on binary outcome variables of dropping out of high school (HS) and receiving income assistance (IA) adjusted for individual-level covariates (age at first pregnancy, 9th grade school performance, birth year, birth order). Model 2 (M2) added household-level covariates (indicator of whether the participant’s mother was a teen mother, indicator of receiving income assistance before high school). Model 3 (M3) added neighborhood-level covariates (average income of people in the neighborhood (in quintiles), urban/rural location, proportion of neighborhood graduated high school). Model 4 (M4) included the fully-adjusted models plus inverse probability weights (IPW) of teen pregnancy calculated based on all covariates stated above. We also ran postestimation Wald tests for equivalency of MEs between key study groups. In Panel A, we show the change in probability of dropping out of high school between Live birth vs. Stillbirth/Miscarriage groups (7.33pp, 95% CI: 1.28, 13.4) and Stillbirth/Miscarriage vs. Abortion groups (0.47pp, 95% CI: -6.53, 5.59). In Panel B, we show the probability change of receiving income assistance between Live birth vs. Stillbirth/Miscarriage groups (20.3, 95% CI: 13.3, 27.3) and Stillbirth/Miscarriage vs. Abortion groups (3.35, 95% CI: -2.30, 9.0)